This pilot parallel group design randomized controlled trial was conducted at Oregon Health & Science University through the Department of Obstetrics & Gynecology between August 2009 and April 2011. A total of 57 participants were enrolled. The trial evaluated the relative efficacy
 of individualized treatment with exercise (control) (n = 22), SMT (n = 15), or NET (n = 20) for pregnancy-related low back pain. The intervention is discussed in detail below. Our primary null hypothesis was that SMT and NET performed no better than exercise for the treatment of PRLBP as measured by the Roland Morris Disability Questionnaire. Our secondary null hypothesis was that SMT and NET performed no better than exercise for the treatment of PRLBP as measured by an 11-point Numeric Pain Rating Scale.
All participants gave oral and written informed consent and received $20 for each study visit, which took about one hour to complete. The study was approved by Oregon Health & Science University’s Institutional Review Board.
Women were eligible for the study if they were pregnant with a singleton and had low back pain of unknown origin that began during pregnancy and was reproducible by manual palpation. Women with health conditions that contraindicated exercise (heart disease, diabetes, thyroid disease, hypertension, Body Mass Index greater than 40, infection, incompetent cervix, bleeding, ruptured membranes, severe anemia, intrauterine growth restriction, poorly controlled seizure disorder, thrombophlebitis, decreased fetal movement, amniotic fluid leakage)
 or manipulation (unrelenting fever, unrelenting night pain, loss of bowel or bladder control, progressive neurological deficit, direct trauma, unexplained weight loss, radiating pain below the knee, cancer, spinal fracture or tumors, blood dyscrasias)
 were excluded from consideration along with women who smoked, consumed alcohol, were medicated with anti-depressants, or had a Roland Morris score above 20 or below 4
. Women were also excluded if they were planning to move during pregnancy, did not read and write English, were not willing to comply with the study visit schedule, or were not willing to be randomized into any of the treatment arms.
Before being randomized participants identified their treatment preference (i.e. exercise, SMT, NET, or none). Treatment preference is a well known predictor of outcomes and thus is a potential confounding variable
[29–31]. The randomization schedule was completed prior to initiating the study and was concealed from all study staff by using consecutively numbered, sealed, opaque envelopes for each strata of preference group. Once a participant was found to be eligible, she would open the consecutive envelope in her preference strata in the presence of the researcher. Participants and the practitioner providing the intervention were not blinded to the intervention after randomization.
Study participants were recruited by fliers in prenatal clinics and by community, magazine, and internet announcements in the Portland, Oregon metropolitan area. Potential participants were first screened by phone and if study eligibility requirements were met, the initial study visit was scheduled. At the initial screening and prior to randomization each treatment group was described so all participants were familiar with all treatments. Participants could enter the study at any point in their pregnancy and remained in the study until they had their baby. All participants continued to receive their usual obstetric or midwifery prenatal care following randomization. The RCT treatment schedule paralleled the prenatal care schedule (once monthly until 28 weeks gestation, twice monthly until 36 weeks gestation, and weekly thereafter).
When the participant arrived for each study treatment visit she completed the primary and secondary outcome measures, drew a pain diagram, and responded to questions about adverse events, treatment by other practitioners, new injuries, and use of medication. After filling out the paperwork all participants were palpated and assessed with standard orthopedic and neurologic exams to ensure they were safe to treat. Each was also palpated to determine if intersegmental dysfunctions were present. Following the assessment each participant was treated in accordance with the protocol for their treatment arm, which is described below.
The lead author served as the therapist and has practiced chiropractic medicine for 15 years. At the baseline visit she screened volunteers for study eligibility through case history and standard orthopedic and neurologic tests. Participants in all three treatment groups were advised to use ice after each treatment since a little soreness is commonly experienced after treatment.
Exercise participants were given a booklet and enthusiastically instructed on pelvic tilts, pelvic floor, gluteus maximus, latissimus dorsi, and hip adductor strengthening exercises to promote low back stability and flexibility. The specific exercises in the booklet have been shown to decrease PRLBP
. The booklet also instructed exercise participants on recommendations for postural and movement patterns that help alleviate low back pain. Finally, warnings in the booklet about when to stop exercising were reviewed with the participant. At each study visit exercises and lifestyle suggestions were reviewed and practiced with participants. Additional individualized stretching or strengthening exercises were prescribed, demonstrated, and practiced at each study visit based on muscle strength and flexibility assessment. Exercises took about 15 min to perform at home and participants were requested to exercise five times weekly.
Spinal Manipulative Therapy
Participants in the SMT group were palpated to determine if each had intersegmental dysfunction prior to manipulating
. Hypomobile joints were isolated through positioning, then a slow force was applied to preload the joint at the physiological end range. After loading the joint, a high velocity, low amplitude thrust was applied to the isolated joint to move it just past the physiological end range in the side posture position for lumbar and sacroiliac lesions. The thrust was applied in the direction, velocity, and amplitude as determined by the clinician from the palpation exam findings
. A hypermobile joint or region was stabilized by creating a fulcrum at a specific joint by the participant lying on padded blocks
. The blocks were always used to adjust a Sacro Occipital Technique Category II pelvis, and the Activator was always used to adjust the pubic symphysis
Neuro emotional technique
Neuro Emotional Technique (NET) is a chiropractic mind-body technique that combines desensitization procedures (such as relaxed breathing and visualization) with elements of Five Element Chinese medicine (such as the association of emotions with certain organs or meridians) and chiropractic medicine (the adjustment of the spinal levels that innervate the organ in question) in an attempt to address cognitive distortions through the use of a semantic algorithm
. The reliability and validity of the manual muscle testing used to guide the technique are unknown
, but muscle testing is considered to be part of the treatment package of NET. The NET standard protocol was followed
At each assessment visit, each participant completed a form indicating if she was following the study protocol, was taking any pain medication, had been injured outside of the study, or had experienced any adverse effects as a result of the intervention. Participants were asked to rate their low back function and pain “today” as measured by the Roland Morris Disability Questionnaire (primary outcome) and an 11-point Numeric Pain Rating Scale (secondary outcome). The Roland Morris Disability Questionnaire is a 24-item index of activities of daily living related to low back function that has been shown to be valid, reliable, and sensitive for measuring changes in mild to moderate disability
[39, 40]. The 11-point Numeric Pain Rating Scale (0–10) has been shown to be valid
. A four point change on the Roland Morris Disability Questionnaire, a two point change on the Numeric Pain Rating Scale, or a 30% change in either from the baseline is considered to be clinically meaningful within-persons
Intention-to-treat analysis was conducted with participants included in the original assigned group. In the primary analysis, linear regression was used to compare the three groups with follow-up disability and pain scores as continuous dependent variables. The covariates included in all between group models were gestational age at entrance into the study, baseline score of the outcome measure, maternal age, and history of low back pain. The endpoint was the eighth study visit assessment or the last study visit if the participant did not complete eight visits.
A secondary responder analysis was conducted for the Roland Morris Disability Questionnaire and Numeric Pain Rating Scale. Two categorical outcome variables were created to reflect a minimum of 50% improvement (yes/no), and a minimum of 30% improvement (yes/no). Differences in the proportion of responders between groups were then calculated. Thirty percent improvement is considered to be the minimal change needed to identify clinically meaningful improvement
[44, 45]. However, 50% improvement is typically the standard used in responder analysis
Missing data were imputed using the last observation carried forward
[48, 49]. Although last observation carried forward method of controlling for missingness is biased toward participants not worsening over the remaining segment of the study, we elected to use this method so our results could be compared with a similarly designed osteopathic trial of treatment for pregnancy related low back pain
. To determine if this approach was appropriate for our data, we also conducted a sensitivity analysis excluding participants with only baseline data.
Statistical significance was set at 0.05 for all tests with no correction for multiple group comparisons. Sample size was limited by available funds and the enrollment window established for this pilot; it was not determined by a priori power analysis. A post hoc power analysis was conducted to identify what effect sizes were detectable with 80% power at the 0.05 level of significance. A study is said to be adequately powered if the detectable effect size is no larger than a clinically important difference between groups. Statistical analysis was performed with SPSS v18.0 (SPSS, Inc, Chicago, IL).